Method for remote intelligent monitoring of tension of stay wire for power grid construction

Abstract

The present application relates to a method for remote intelligent monitoring of tensioned wire force for power grid construction. In view of the difficulties in state tracing and abnormality determination during installation and use of tensioned wires, by collecting installation process parameters, environmental exposure trajectory, tension time sequence characteristics and equipment health attenuation curve, a unique service state fingerprint ID is generated using hash coding, and a cross-dimension function coupling graph is constructed to realize dynamic labeling of the function role of the tensioned wire. According to the characteristics of the main load type, the redundant check type and the transition anchoring type, the double-mode envelope threshold, the deviation consistency threshold and the gradient sensitive trigger rule are set respectively to make graded early warning judgment on real-time tension data. Through feedback closed-loop optimization of threshold parameter interval, the intelligence and reliability of tensioned wire monitoring are improved.

Description

Technical Field

[0001] This invention relates to the field of power grid construction monitoring and hierarchical early warning technology, and in particular to a remote intelligent monitoring method for guy wire stress in power grid construction. Background Technology

[0002] In the current field of remote intelligent monitoring of power grid construction and guy wire stress, mainstream technologies mostly adopt centralized data acquisition and static threshold judgment strategies. Traditional solutions collect data from the tension sensors of various monitoring devices, push it uniformly to the server, and trigger warnings for all guy wires based on manually set tension warning values ​​or empirical formulas. Some systems attempt to introduce multi-source data fusion and risk weight mapping technologies to form a hierarchical alarm model by weighting and integrating the guy wire's load-bearing capacity, installation parameters, or environmental factors. However, most of these existing technologies are still at the static modeling stage centered on the equipment, only achieving basic anomaly detection and alarm priority allocation, and failing to fully consider the impact of changes in the functional role of guy wires and the dynamic evolution of their service status during the power grid construction cycle on risk response.

[0003] Some existing technologies attempt to optimize alarm grading through weight presets, fusion modeling, or communication scheduling to adjust push priorities in multi-device scenarios. However, they struggle to dynamically map the functional roles and health status of the guy wire itself in actual engineering projects. The failure to establish an adaptive grading and early warning mechanism driven by the guy wire's service status limits the system's intelligence and prevents it from achieving "task-centric" risk management. The technical field addressed by this invention urgently needs to overcome the limitations of static device attribute discrimination and establish a dynamic threshold generation and grading early warning method based on guy wire service status fingerprints and functional role hierarchies. This would achieve a fundamental leap in early warning logic from static unified values ​​to role-adaptive ranges. Summary of the Invention

[0004] This application provides a remote intelligent monitoring method for guy wire stress in power grid construction, which aims to solve one of the problems or issues of the prior art mentioned in the background section.

[0005] The remote intelligent monitoring method for guy wire stress in power grid construction provided in this application specifically includes: S1: Obtain four types of initial fingerprint elements solidified during the installation phase of the cable monitoring device, including installation process parameters, environmental exposure trajectory, tension timing characteristics, and equipment health decay curve, and perform edge hash encoding on the four types of initial fingerprint elements to generate an irreversibly mapped cable service status fingerprint ID.

[0006] S2: Construct a fingerprint evolution graph, map the service status fingerprint ID of the guy wires to fingerprint evolution graph nodes, and calculate the cross-dimensional similarity between different nodes based on material grade, altitude and ultraviolet accumulation, so as to generate graph edge weights that characterize the functional coupling relationship between guy wires.

[0007] S3: Utilize the edge weights of the graph to perform subgraph aggregation operation on the fingerprint evolution graph, identify the wire clusters with strong functional coupling, and combine the topological position, historical tension peak distribution dispersion and redundancy check success rate indicators to dynamically label each node in the cluster with functional role labels of main load-bearing type, redundancy check type or transition anchoring type.

[0008] S4: For the functional role labels marked as main load-bearing type, extract the theoretical maximum tension and the slope of tension change in the past ten minutes under the current working condition, and generate a dual-modal envelope threshold range for main load-bearing type that includes the upper limit of dynamic safety factor and the lower limit of instability prevention.

[0009] S5: For functional role tags marked as redundant verification type, select the real-time tension data of the main load-bearing type guy wire in the same cluster as the benchmark reference, calculate the duration of continuous deviation of the ratio of the current guy wire tension to the benchmark reference, and generate a redundant verification type deviation consistency threshold condition for determining consistency anomalies.

[0010] S6: For functional roles labeled as transitional anchoring, a millisecond-level response window is set to capture the rate of tension rise, and combined with the steady-state tension tolerance parameter, a gradient-sensitive threshold triggering rule that is sensitive to transient impacts is generated.

[0011] S7: Input the real-time collected tension data into the corresponding functional role tag-bound main load-bearing dual-modal envelope threshold range, redundant verification type deviation consistency threshold condition, or gradient sensitive threshold triggering rule for matching and judgment, so as to generate graded early warning triggering signals.

[0012] S8: Based on the false alarm or missed alarm feedback record of the graded early warning trigger signal, perform closed-loop fine-tuning operation on the dynamic safety coefficient, ratio deviation duration or response window parameter on which the signal is generated, so as to update the role dynamic range parameter configuration for the next cycle.

[0013] The remote intelligent monitoring method for guy wire stress in power grid construction provided in this application has the following beneficial effects: (1) The guy wire status monitoring and adaptive early warning method provided in this application effectively overcomes the problems of delayed early warning and frequent misjudgments caused by relying on manually set thresholds, static classification models or isolated data sources in traditional monitoring systems. Existing technologies usually use sensor data directly for threshold comparison or multi-source fusion analysis, which lacks in-depth characterization of individual differences and service background of guy wires and is difficult to adapt to the changing needs of dynamic working conditions under complex construction conditions. However, this solution realizes the leap from "single equipment monitoring" to "group behavior modeling" by constructing a fingerprint evolution map with engineering semantic correlation and dynamically calculating the edge weight between nodes with cross-dimensional similarity. This atlas does not rely on preset cluster centers or fixed weight allocation. Instead, it automatically identifies strongly correlated string clusters based on multi-dimensional coupling relationships such as material, environment, and historical response. This significantly improves the accuracy and robustness of abnormal pattern recognition under similar working conditions. In particular, it can maintain a stable cluster structure when facing the application of new materials, non-standard construction scenarios, or sudden local climate changes, thus providing a reliable topological foundation for subsequent differentiated early warning strategies.

[0014] (2) Based on the functional guy wire clusters identified by the fingerprint evolution map, the system further introduces a dynamic role labeling mechanism. According to the three indicators of topological position, tension distribution dispersion and verification success rate, each guy wire is divided into three functional roles in real time: "main load-bearing type", "redundant verification type" or "transitional anchoring type". Based on this, a differentiated threshold generation logic is implemented, which completely gets rid of the sensitivity imbalance problem caused by the unified alarm rules in the traditional method. For the main load-bearing type guy wire, a dual-modal envelope threshold is adopted to take into account the theoretical safety boundary and the real-time change trend, so as to prevent false alarms caused by excessively tight constraints or instability risks caused by excessively loose tolerance. The redundant verification type guy wire focuses on the deviation consistency judgment and only triggers an alarm when its tension ratio with the main load-bearing type guy wire deviates continuously, which greatly reduces the false alarm rate under environmental noise interference. The transitional anchoring type guy wire is configured with a gradient sensitive threshold, which relaxes the steady-state tolerance range while responding quickly to transient impacts, and fully adapts to its temporary bearing characteristics. More importantly, all thresholds are adaptively adjusted through a closed-loop feedback mechanism based on historical false alarms / missed alarms. For example, if there are continuous false alarms, the safety factor is reduced, and if no effective comparison is conducted for a long period of time, a role downgrade reminder is triggered. This forms a complete control loop of "monitoring-decision-feedback-optimization", which significantly enhances the system's self-evolution capability and on-site adaptability. It can achieve long-term stable operation without frequent manual intervention or expert experience parameter tuning.

[0015] The aforementioned technical approaches collectively construct an intelligent monitoring system driven by the evolution of functional roles, achieving a paradigm shift in early warning logic from "equipment-centric" to "task-centric." Compared to traditional approaches relying on multi-source fusion modeling, risk level mapping, or communication link scheduling, this solution is more aligned with the actual engineering progress pace. It can continuously update the map structure and dynamically adjust role assignments during the rolling construction of overhead line sections, exhibiting excellent scalability and engineering feasibility. Furthermore, by employing hash encoding throughout to protect the privacy of original data and ensuring completely decentralized threshold generation, the system significantly improves response sensitivity and decision-making accuracy while guaranteeing information security. It is particularly suitable for high-reliability scenarios such as large-scale transmission line construction, wiring in complex mountainous terrains, and operation and maintenance in extreme climate areas, providing practical technical support for the safe and intelligent upgrading of power infrastructure. Attached Figure Description

[0016] Figure 1 This is the main flowchart of a remote intelligent monitoring method for guy wire stress in power grid construction.

[0017] Figure 2 This is a sub-flowchart of a method for remote intelligent monitoring of guy wire stress in power grid construction.

[0018] Figure 3 This is another sub-flowchart of the remote intelligent monitoring method for guy wire stress in power grid construction. Detailed Implementation

[0019] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0020] The following disclosure provides many different embodiments or examples for implementing different structures of the invention. To simplify the disclosure, specific examples of components and arrangements are described below. Of course, these are merely examples and are not intended to limit the invention. Furthermore, reference numerals and / or letters may be repeated in different examples; such repetition is for simplification and clarity and does not in itself indicate a relationship between the various embodiments and / or arrangements discussed.

[0021] like Figure 1 As shown, this application provides a remote intelligent monitoring method for guy wire stress in power grid construction, specifically including: S1: Obtain four types of initial fingerprint elements solidified during the installation phase of the cable monitoring device, including installation process parameters, environmental exposure trajectory, tension timing characteristics, and equipment health decay curve, and perform edge hash encoding on the four types of initial fingerprint elements to generate an irreversibly mapped cable service status fingerprint ID.

[0022] S2: Construct a fingerprint evolution graph, map the service status fingerprint ID of the guy wires to fingerprint evolution graph nodes, and calculate the cross-dimensional similarity between different nodes based on material grade, altitude and ultraviolet accumulation, so as to generate graph edge weights that characterize the functional coupling relationship between guy wires.

[0023] S3: Utilize the edge weights of the graph to perform subgraph aggregation operation on the fingerprint evolution graph, identify the wire clusters with strong functional coupling, and combine the topological position, historical tension peak distribution dispersion and redundancy check success rate indicators to dynamically label each node in the cluster with functional role labels of main load-bearing type, redundancy check type or transition anchoring type.

[0024] S4: For the functional role labels marked as main load-bearing type, extract the theoretical maximum tension and the slope of tension change in the past ten minutes under the current working condition, and generate a dual-modal envelope threshold range for main load-bearing type that includes the upper limit of dynamic safety factor and the lower limit of instability prevention.

[0025] S5: For functional role tags marked as redundant verification type, select the real-time tension data of the main load-bearing type guy wire in the same cluster as the benchmark reference, calculate the duration of continuous deviation of the ratio of the current guy wire tension to the benchmark reference, and generate a redundant verification type deviation consistency threshold condition for determining consistency anomalies.

[0026] S6: For functional roles labeled as transitional anchoring, a millisecond-level response window is set to capture the rate of tension rise, and combined with the steady-state tension tolerance parameter, a gradient-sensitive threshold triggering rule that is sensitive to transient impacts is generated.

[0027] S7: Input the real-time collected tension data into the corresponding functional role tag-bound main load-bearing dual-modal envelope threshold range, redundant verification type deviation consistency threshold condition, or gradient sensitive threshold triggering rule for matching and judgment, so as to generate graded early warning triggering signals.

[0028] S8: Based on the false alarm or missed alarm feedback record of the graded early warning trigger signal, perform closed-loop fine-tuning operation on the dynamic safety coefficient, ratio deviation duration or response window parameter on which the signal is generated, so as to update the role dynamic range parameter configuration for the next cycle.

[0029] Step S1: Obtain four types of initial fingerprint elements solidified during the installation phase of the cable monitoring device, including installation process parameters, environmental exposure trajectory, tension timing characteristics, and equipment health decay curve. Then, perform edge-end hash encoding on these four types of initial fingerprint elements to generate an irreversibly mapped cable service status fingerprint ID. Specifically, this includes: S1.1: Obtain static configuration data and real-time sampling sequences from the built-in storage unit of the guy wire monitoring device, and perform standardized analysis on the installation process parameters, including guy wire material grade, diameter specification, preload setting value, anchoring angle and connection hardware type, to generate a structured installation process feature vector.

[0030] The static configuration data and real-time sampling sequence of the sensor, stored in the built-in storage unit of the guy wire monitoring device, are used as input conditions. The initialization parameter segment in the non-volatile memory chip is then used to parse the raw encoded data of the guy wire material grade, diameter specification, preload setting, anchoring angle, and connection hardware type. Type matching mapping processing is performed on the raw encoded data, converting the material grade into material property codes, the diameter specification into geometric dimension vectors, the preload setting into mechanical reference parameters, the anchoring angle into spatial attitude parameters, and the connection hardware type into structural connection attributes. Based on the different physical dimensions of each parameter, parameter normalization processing is performed, mapping the mechanical, geometric, and spatial parameters to a dimensionless feature space according to a unified dimensional rule. A multi-dimensional feature encoding function is used to combine the normalized parameters to form a structured installation process feature vector. A matrix splicing operation is then used to arrange the material property codes, geometric dimension vectors, mechanical reference parameters, spatial attitude parameters, and structural connection attributes into a multi-dimensional vector structure with a fixed order.

[0031] By using normalized mapping and encoding combination processing, the static configuration data and real-time sampling sequence from the previous step are transformed into structured installation process feature vectors with unified dimensions that can be used for subsequent spatiotemporal benchmark establishment and feature fusion, thereby achieving standardization and reusability of installation parameters.

[0032] S1.2: Based on the installation process feature vector as a spatiotemporal reference, the data stream of the micro meteorological sensor integrated in the photovoltaic power supply module and the positioning information of the 4G communication module are called to perform spatiotemporal alignment and cumulative calculation of the environmental exposure trajectory of the cable's altitude, daily average ultraviolet intensity, cumulative rainfall duration and temperature fluctuation frequency, so as to generate a multidimensional environmental exposure feature sequence.

[0033] Using the input installation process feature vector as a unified reference benchmark for spatial location and time series, the data stream of the micro meteorological sensor integrated in the photovoltaic power supply module is invoked to read the raw time-series signal containing continuous observations such as light intensity, ultraviolet irradiance, air temperature, and rainfall detection. Simultaneously, the positioning information from the 4G communication module is retrieved to obtain the latitude, longitude coordinates, and altitude values ​​at the corresponding time. The positioning information and meteorological sensor data are synchronized to achieve a precise correspondence between altitude and the timestamps of meteorological observations. Noise reduction and stabilization filtering are performed on the synchronized altitude data to eliminate the interference of instantaneous positioning fluctuations on environmental feature estimation. Based on the synchronized ultraviolet irradiance data, the daily average ultraviolet intensity is calculated using integral accumulation. The number of days in the construction cycle is used as the dividing window, and the ultraviolet intensity within each window is mathematically expected to form a daily distributed ultraviolet intensity sequence. In the rainfall detection data, threshold judgment logic is used to identify rainfall events, and the duration of rainfall events throughout the entire construction cycle is statistically calculated and accumulated to obtain a cumulative rainfall duration index. Differential operations are performed on the temperature data to obtain the temperature change amplitude within each sampling period. The amplitude value is compared with a preset temperature difference threshold to identify events where temperature fluctuations exceed the threshold, and the frequency of such events is counted to generate a temperature difference fluctuation frequency index. By spatiotemporally aligning the above index with the original altitude data, a multidimensional environmental exposure feature sequence is constructed, including altitude, daily average ultraviolet radiation intensity, cumulative rainfall duration, and temperature difference fluctuation frequency.

[0034] By using the above processing method, the installation process feature vector of the previous step is transformed into a quantifiable and comparable set of environmental exposure features, so as to achieve accurate constraint and normalization analysis of the working environment when extracting tension time sequence features in the subsequent process.

[0035] S1.3: Using the multidimensional environmental exposure feature sequence as the working condition constraint, select the tension raw data window of the 72-hour unloaded static period before construction, and use digital signal processing to extract the tension mean drift rate, low frequency oscillation energy ratio and transient impact response decay time constant to generate a tension time series feature spectrum characterizing mechanical behavior.

[0036] S1.4: Combining the tension time-series characteristic spectrum as the health assessment input, analyze the power spectral density distribution of the ADC sampling noise inside the MCU main control chip and the signal strength attenuation trend curve of the 4G module, and perform equipment aging characterization modeling processing to generate an equipment health attenuation curve that reflects the performance degradation of the monitoring terminal itself.

[0037] A health correlation analysis was performed on the stress fluctuation index contained in the tension time series characteristic spectrum to determine the power spectral density curve of the noise signal during the stress sampling process of the monitoring terminal as the first input data source.

[0038] A fast Fourier transform operation is performed on the sampling sequence of the ADC inside the MCU main control chip to extract the frequency domain energy distribution, and the difference in the shape of the power spectral density curve is calculated in the low-frequency to high-frequency range.

[0039] The power spectral density curve is differentially calculated with the reference noise-free signal to form a noise energy ratio sequence. The noise fundamental frequency peak and broadband noise energy ratio are then extracted as health attenuation criteria.

[0040] The signal strength of the 4G communication module in different time windows is processed by moving average to form a signal strength attenuation trend curve, and the trend slope and curve fluctuation amplitude are calculated as the second input data source.

[0041] Based on the first and second input data sources, a device health aging determination function is constructed, and the comprehensive aging coefficient is calculated using the following formula: ; Where α is the noise sensitivity weight, P is the noise energy percentage, β is the signal attenuation weight, and S is the signal strength attenuation slope.

[0042] The overall aging coefficient is fitted with the aging coefficient sequence formed by each sampling period over time to output the equipment health degradation curve of the monitoring terminal from installation to the current stage.

[0043] This processing method transforms the tension time-series characteristic spectrum and communication link attenuation characteristics into a single health attenuation curve that characterizes the degree of equipment performance degradation, enabling early quantitative assessment of the risk to the long-term stable operation of monitoring terminals.

[0044] For example, at a high-voltage power line construction site, the ADC sampling frequency in the MCU main control chip is configured to 2000 Hz, and the peak power spectral density is 0.015 W / Hz, with a broadband noise energy ratio P of 0.22. The 4G module's signal strength has attenuated from -65 dBm to -72 dBm over the past 7 days, with a trend slope S of -1 dBm / day. The noise sensitivity weight α is set to 0.6, and the signal attenuation weight β is set to 0.4. Substituting these values ​​into the formula: ; The overall aging coefficient K is -0.316, indicating a significant decline in equipment performance. As this coefficient continues to decrease, the health degradation curve shows a sharp negative increase, suggesting that the equipment maintenance team should perform on-site module replacement within 5 days to ensure the stability of the tension monitoring and early warning functions.

[0045] S1.5: Integrate the installation process feature vector, multidimensional environmental exposure feature sequence, tension time series feature spectrum and equipment health decay curve to form four types of initial fingerprint element sets. Use edge hash encoding to perform irreversible mapping compression processing on the four types of initial fingerprint element sets to generate a unique and fixed wire service status fingerprint ID.

[0046] Step S2: Construct a fingerprint evolution graph, mapping the service status fingerprint IDs of the guy wires to nodes in the fingerprint evolution graph, and calculating the cross-dimensional similarity between different nodes based on material grade, altitude, and UV accumulation to generate graph edge weights representing the functional coupling relationship between guy wires. Specifically, this includes: S2.1: Obtain the irreversible mapped service status fingerprint ID of the guy wire generated in the previous steps, and the four types of initial fingerprint elements it contains: installation process parameters, environmental exposure trajectory, tension time sequence characteristics, and equipment health decay curve. Perform structured parsing processing on the four types of initial fingerprint elements to extract the material grade scalar, altitude value, and ultraviolet cumulative amount statistics for similarity calculation as the basic input data for cross-dimensional similarity calculation.

[0047] S2.2: Based on the service status fingerprint ID of the guy wire in the basic input data, construct a fingerprint evolution map, map each unique service status fingerprint ID of the guy wire to an independent map node in the fingerprint evolution map, and associate and store the corresponding material grade scalar, altitude value and ultraviolet cumulative statistical value to generate a set of fingerprint evolution map nodes with complete attribute description.

[0048] Based on the service status fingerprint IDs of the guy wires in the basic input data, a fingerprint evolution map is constructed. The node instantiation module is called to index and register the unique fingerprint IDs, mapping each ID to an independent node unit in the fingerprint evolution map, and allocating storage space for each node unit to hold associated attributes. For each node unit, the material grade scalar obtained from structured parsing is bound sequentially, converted into a calculable attribute value through a material coding mapping table, and written into the node attribute field. The parsed altitude values ​​are standardized in units and adjusted in precision, stored as the node's geographic environment attribute, and combined with spatial positioning coordinate indexing to achieve fast retrieval. Time-series normalization is performed on the cumulative ultraviolet radiation statistics, and the normalization result is stored as a climate exposure attribute in the node, establishing an attribute sub-table associated with meteorological factors. An attribute binding verification mechanism is used to check the completeness and type consistency of the material, altitude, and ultraviolet radiation attributes of each node, ensuring that each node in the node set has a complete and reliable attribute description. Through continuous processing of node instantiation and attribute binding, the basic input data of the previous step is transformed into a set of fingerprint evolution map nodes with comprehensive material, geographical and climate attributes, thereby providing an accurate and structured node data source for subsequent cross-dimensional similarity calculations.

[0049] S2.3: Traverse any two different fingerprint evolution map nodes in the fingerprint evolution map node set, extract the material grade scalar, altitude value and ultraviolet cumulative statistical value corresponding to the two nodes respectively, and use the preset cross-dimensional weighted similarity measurement method to calculate the discrete matching degree of the material grade scalar, calculate the normalized distance inverse ratio of the altitude value, and calculate the trend consistency correlation coefficient of the ultraviolet cumulative statistical value, so as to generate the material matching factor, elevation proximity factor and irradiation convergence factor that characterize the single-dimensional similarity between the two nodes.

[0050] For any two different nodes in the fingerprint evolution graph node set, call the node attribute parsing interface to extract their respective material grade scalar, altitude value, and ultraviolet cumulative amount statistics as initial input data.

[0051] Input the material grade scalar into the discrete matching degree calculation module. Perform a one-to-one complete matching judgment based on the material code lookup table or calculate the matching weight through similar material category mapping. The matching degree output value is limited to the numerical range of 0 to 1.

[0052] The altitude value is input into the normalized distance inverse ratio calculation module. First, the absolute value of the altitude difference is calculated, then normalization is performed based on the local maximum altitude difference, and then the elevation proximity factor is generated by mapping through the inverse ratio function to ensure that a smaller absolute altitude difference corresponds to a higher factor value.

[0053] Input the cumulative ultraviolet radiation statistics into the trend consistency correlation coefficient calculation module. First, generate the ultraviolet radiation cumulative vector sequence of two nodes in the same time span. Then, perform mean centering and variance standardization. Calculate the irradiance convergence factor using the Pearson correlation coefficient formula.

[0054] The material matching factor, elevation proximity factor, and irradiation convergence factor are output in the order of execution to form a complete set of indicators representing the single-dimensional similarity between two nodes. The chain operation ensures that the calculation of each factor depends on the input attributes of the previous step, thereby realizing the decomposition calculation of multi-dimensional feature similarity assessment.

[0055] By independently generating factors in the three domains of material, geographical environment, and climate exposure, clear and quantifiable single-dimensional original evaluation values ​​can be provided for subsequent multi-source feature fusion calculations, thereby improving the accuracy and stability of cross-dimensional similarity calculations.

[0056] S2.4: Based on the generated material matching factor, elevation proximity factor, and irradiation convergence factor, perform multi-source feature fusion operation, and aggregate the material matching factor, elevation proximity factor, and irradiation convergence factor into a single comprehensive coupling strength value through linear weighted summation, so as to generate a cross-dimensional similarity score that characterizes the overall functional coupling degree between the two lines in terms of material properties, geographical environment, and climate exposure.

[0057] Based on the material matching factor, elevation proximity factor, and irradiation convergence factor generated in the previous steps as input conditions for multi-source fusion calculation, the weight allocation strategy for each factor depends on the working condition importance coefficient and historical coupling degree verification records.

[0058] Based on the weight allocation strategy, the material matching factor is used as the original matching degree weight term, the elevation proximity factor is used as the geographical environment weight term, and the irradiation convergence factor is used as the climate exposure weight term to construct a multi-source linear fusion formula.

[0059] The three types of weight terms are input into the linear weighted summation unit, and the numerical processing of summation after multiplying the terms is performed to output the comprehensive coupling strength value.

[0060] The comprehensive coupling strength values ​​were standardized to zero mean and unit variance using a normalization method to eliminate the influence between different units.

[0061] A threshold pruning process is performed on the standardized comprehensive coupling strength values. Results below the set threshold are identified as weak coupling relationships, while results above the set threshold are retained as strong coupling candidates.

[0062] Through the above fusion and cropping, a cross-dimensional similarity score is obtained to characterize the overall functional coupling degree of the two lines in terms of material properties, geographical environment and climate exposure.

[0063] By using multi-source feature fusion and normalization, the single-dimensional similarity index generated in the previous step is transformed into a cross-dimensional similarity score with a unified scale, thereby quantifying and making comparable the degree of functional coupling between different pull lines.

[0064] S2.5: Based on the generated cross-dimensional similarity score, establish a directed or undirected connection between the corresponding two fingerprint evolution graph nodes, and assign the cross-dimensional similarity score to the connection as the graph edge weight to generate a weighted fingerprint evolution graph structure that fully represents the strength of functional coupling between the lines, for subsequent subgraph agglomeration operations.

[0065] Given cross-dimensional similarity scores as input, the connection relationship instantiation process is performed on fingerprint evolution graph nodes corresponding to any two different lines, binding node indices and score values ​​as core attributes of graph connection entities. For the attribute binding of score values, an edge weight allocation function is called to achieve precise assignment of scores to connection entities through scalar propagation. Simultaneously, a directed or undirected connection type marker field is established, with its Boolean value determined by the functional coupling directionality determination result from the previous step. For directed connection types, in-degree correlation mapping is performed, synchronously updating the out-degree count of the source node and the in-degree count of the target node, thereby completing the topological integrity of the evolution graph in directional coupling scenarios. For undirected connection types, bidirectional adjacency matrix filling is performed to ensure that nodes in either direction can be accessed during traversal, guaranteeing global connectivity for subsequent subgraph aggregation. All sets of entities with established connections are stored in a structured manner, using sparse matrix encoding to compress node indices, connection type marker fields, and edge weight values ​​into matrix triples, thus forming a weighted fingerprint evolution graph structure with clearly defined directional attributes. Through the above processing method, the cross-dimensional similarity score generated by S2.4 is transformed into weighted graph edge data representing the strong and weak relationships of functional coupling, realizing the direct callability and performance optimization of the fingerprint evolution graph in subsequent condensed clustering calculation.

[0066] like Figure 2 As shown, step S3 involves using the edge weights of the graph to perform a subgraph agglomeration operation on the fingerprint evolution graph, identifying wire clusters with strong functional coupling, and dynamically labeling each node within the cluster with functional roles such as main load-bearing type, redundant check type, or transition anchoring type, based on topological location, historical tension peak distribution dispersion, and redundancy check success rate. Specifically, this includes: S3.1: Obtain the set of graph edge weights representing the functional coupling relationship between the wires generated in the previous steps, and perform iterative cutting processing on the fingerprint evolution graph to generate an initial set of wire clusters with strong internal connectivity and weak external correlation.

[0067] Based on the edge weight set representing the functional coupling relationship between the lines generated in the previous steps, this edge weight set is used as the input condition for modular density analysis. A preset community partitioning iterative framework is invoked to initialize the segmentation state matrix of the fingerprint evolution graph. For this segmentation state matrix, initial value calculation processing of the modular density objective function is performed. The objective function has the following form: ; Among them, A ij Let k be the edge weight between nodes i and j. i Let k be the degree of node i. j Let m be the degree of node j, m be half the total weight of the entire graph, δ be the community consensus factor, and c be the degree of node j. i With c j These are the cluster identifiers for the two nodes. Based on the objective function calculation result, all node movement strategies that can improve the Q-value are retrieved, and neighborhood swapping is performed on nodes that meet the improvement conditions, temporarily migrating the nodes to clusters with higher weight connections. After completing one neighborhood swap, the new cluster partitioning result is evaluated using modular density gain. When the gain value is higher than a preset convergence threshold, the node migration and cluster merging operations continue iteratively. During the iteration process, fast recalculation is performed on the edge weight updates generated by each cluster merge to ensure that the input matrix of the modular density objective function is synchronized in real time. All iterations terminate when the improvement of the objective function Q-value is lower than the minimum gain threshold and no effective migrations are generated consecutively, outputting an initial set of wired clusters with strong internal connectivity and weak external correlation.

[0068] By using the modular density maximization iterative cutting process described above, the cross-dimensional similarity edge weight results from the previous step are transformed into an initial cluster set that can be used for topological importance assessment and functional role labeling, thereby achieving accurate identification of high coupling within clusters and isolation of weak connections between clusters.

[0069] For example, in a 220kV transmission line construction monitoring scenario, the fingerprint evolution map of the guy wire service status contains 48 nodes, with a 48×48 dimension edge weight matrix, edge weights ranging from 0.12 to 0.98, and an average degree of 14. The modular density maximization method is used with a convergence threshold of 0.001, a minimum gain threshold of 0.0005, and an initial Q value of 0.42. In the first iteration, 17 node migrations are detected, which improves the Q value; after neighborhood swapping, the Q value rises to 0.53. In the second iteration, only 8 nodes migrate, resulting in a gain of 0.006 and a Q value of 0.536. In the third iteration, the gain drops to 0.0004, triggering the termination condition. The output initial cluster set contains 5 clusters, with an average intra-cluster edge weight of 0.84 and an average inter-cluster edge weight of 0.21. The cluster partitioning results, verified by subsequent topology analysis, significantly improve the accuracy of distinguishing between critical stressed guy wires and redundant guy wires, ensuring the rationality of early warning priority allocation.

[0070] S3.2: Extract the construction traction process topology location data of each node in the initial guy wire cluster set, and perform connectivity component analysis processing in combination with the physical connection adjacency matrix between members within the cluster to generate a topology hierarchy index that reflects the relative importance of the guy wire in the force transmission path.

[0071] The construction traction process topology location data of each node in the initial guy wire cluster set is retrieved, and the absolute position number and relative hierarchical position index of each node in the physical force path are extracted as basic inputs. The extracted topology location data is then checked against the physical connection adjacency matrix among cluster members to ensure consistency between the node index and the matrix row and column correspondence. Based on the consistency-checked adjacency matrix, connected component partitioning is performed, and a depth-first traversal method is used to identify the set of reachable paths between any two nodes in the matrix. For each set of nodes in a connected component, the sum of the force transmission path lengths from that node to all other nodes in the cluster is calculated, and the path lengths are normalized to generate a node centrality index. The node centrality index is combined with the topology location index and fused into a topology hierarchical score using a weighted summation method. The weight coefficients are set according to the importance level of the location in the construction traction process. A hierarchical sorting mechanism is used to arrange the nodes in the cluster from high to low topology hierarchical scores, generating a topology hierarchical index to characterize the relative importance of the guy wire in the force transmission path. By combining connected component analysis with topological hierarchy fusion processing, the initial wire cluster data from the previous step is transformed into hierarchical index data with topological importance ranking capabilities, thus enabling structured input for subsequent comprehensive performance scoring of stability and reliability.

[0072] For example, in a high-voltage transmission line erection project, the initial guy wire cluster set contains 6 nodes, numbered 1 to 6 in the topology. Nodes 1, 2, and 3 are located in the main load-bearing direction of the tower, while nodes 4, 5, and 6 are located in the transition anchorage direction. The adjacency matrix is ​​6×6 in size, with a value of 1 indicating a direct physical connection and a value of 0 indicating no connection. After checking the row-column correspondence between the adjacency matrix and the position numbers, a depth-first traversal is performed to obtain two connected components: connected component A contains nodes 1, 2, 3, and 4, and connected component B contains nodes 5 and 6. For nodes within connected component A, the sum of the shortest path lengths to other nodes is calculated. For example, the sum of the path lengths for node 1 is 4, for node 2 it is 3, for node 3 it is 5, and for node 4 it is 6. These sums are normalized to the [0,1] interval, yielding centrality indices of 0.66, 0.5, 0.83, and 1.0, respectively. The centrality metric and the position index are fused, with a weighting coefficient of 0.7 for position importance and 0.3 for centrality. For example, the fused score for node 1 is 0.7×1+0.3×0.66 =0.898, and the score for node 2 is 0.7×0.8+0.3×0.5 =0.71. Nodes within the cluster are arranged from highest to lowest score, resulting in a topological hierarchy index order of 1, 3, 2, 4, 5, 6. This index is used for subsequent comprehensive performance scoring calculations, significantly improving the accuracy of dynamic role labeling.

[0073] S3.3: Retrieve the historical tension peak distribution dispersion statistics and redundancy check success rate records of each initial tension cluster, and perform multi-dimensional feature fusion calculation on the topology hierarchical index to generate a comprehensive performance score that characterizes the stability and reliability within the cluster.

[0074] Historical tension peak distribution dispersion statistics and redundancy check success rate records of the initial tension cluster set are retrieved as the basic inputs for multi-dimensional feature fusion. Normalization of the dispersion statistics is performed to unify and calibrate the inconsistency in peak distribution scale caused by differences in sampling ranges between different clusters. Based on the normalized peak distribution dispersion sequence, a weighted entropy calculation expression is constructed, using dispersion as an input term for the peak probability distribution, and entropy analysis is performed to quantify the stability of tension peaks within a cluster. During the normalized dispersion input entropy analysis, redundancy check success rate records are simultaneously introduced as the second dimension feature, and a logarithmic transformation is performed on them to reduce the nonlinear impact of extremely high or low success rates on model weight allocation. The above two dimensions are multiplied by weight coefficients associated with the cluster topology hierarchy index, which are calculated based on the relative position of nodes within the cluster in the force transmission chain and cross-cluster connectivity. The comprehensive performance score is calculated using the following formula: ; Where S is the overall performance score, w1 and w2 are the weighting coefficients bound to the topology level index, and p iLet be the probability value of the i-th peak distribution interval, and R be the number of successful redundancy checks. In the formula, the entropy term reflects the stability of the peak distribution, and the redundancy check term reflects the reliability of the cluster. The weighted sum of these two terms forms the comprehensive score. After calculation, the comprehensive performance score sequence is mapped back to the corresponding string cluster, achieving a quantitative expression of intra-cluster stability and reliability.

[0075] Through the above-mentioned multi-dimensional feature fusion processing method, the topological hierarchical index generated in the previous step, along with the historical tension peak dispersion and redundancy check success rate records, are transformed into a comprehensive performance score that characterizes the stability and reliability within the cluster, thus realizing the accurate sorting basis before functional role division.

[0076] For example, in a high-voltage transmission line construction scenario, an initial cluster consisting of six guy wires is analyzed. The peak distribution dispersion statistic, after normalization, ranges from 0.15 to 0.32. The original redundancy check success rate is between 20 and 45, and after logarithmic transformation, it ranges from 3.04 to 3.83. Weighting coefficients w1 and w2 are set to 0.6 and 0.4, respectively, and w1 is adjusted to 0.75 based on the location of a main load-bearing node within the cluster on the core load-bearing path. In the entropy calculation, p is taken as the row vector of each guy wire in the peak distribution probability matrix, and the entropy result is between 1.12 and 1.35. The redundancy check term avoids zero values ​​by adding 1 to the denominator in the formula, and the calculated result is between 1.52 and 1.57. After substituting into the formula, the overall performance score S ranges from 1.33 to 1.42. It is clear that high-scoring nodes are candidates for main load-bearing capacity with high stability and reliability, while low-scoring nodes are transitional types that are prone to degradation. The final output of the overall performance score effectively improves the accuracy of role classification.

[0077] S3.4: Based on the ranking results of the comprehensive performance scores, perform hierarchical mapping matching processing on the nodes in the initial wire cluster set. Nodes with high scores and located in key topological positions are mapped to main load-bearing functional roles, nodes with medium scores and comparable value are mapped to redundancy verification functional roles, and nodes with low scores and located at the edge are mapped to transitional anchoring functional roles, so as to generate a dynamic wire cluster structure with functional role labels.

[0078] When performing functional role hierarchical mapping on the comprehensive performance score ranking results generated in the previous steps, the comprehensive performance score dataset is used as input. A multi-dimensional mapping judgment model is constructed to classify the roles of each node by combining the location information, physical connection relationship, and functional importance index of each node in the guy wire cluster topology. Elements with scores higher than a preset high threshold and located at key nodes in the force transmission path are input to the main load-bearing role judgment unit, and are assigned a main load-bearing functional role label based on their load-bearing capacity and stability requirements. Elements with scores between the high and low thresholds and capable of effectively comparing tension changes in main load-bearing nodes are input to the redundancy verification role judgment unit, and are assigned a redundancy verification functional role label based on their comparison capability and historical redundancy verification success rate. Elements with scores lower than a preset low threshold and located at the edge of the force path, mainly used for short-term anchoring or transitional connections, are input to the transitional anchoring role judgment unit, and are assigned a transitional anchoring functional role label based on their transient force response characteristics. Each tag binding operation adds a role attribute field to the node data structure and generates a dynamic string cluster structure output containing node ID, role tag, and rating value.

[0079] By using hierarchical matching processing based on comprehensive performance scores, the agglomerative subgraph structure from the previous step is transformed into functional role labeling data that can be used to generate differentiated thresholds, thereby achieving adaptive optimization of early warning logic for different pull-line nodes.

[0080] For example, in the construction scenario of a 220kV transmission line on the Qinghai-Tibet Plateau, an initial guy wire cluster contains 8 nodes with comprehensive performance scores of 82.5, 79.3, 65.8, 61.4, 55.2, 48.7, 44.1, and 39.9, respectively. The high threshold is set to 80, and the low threshold is set to 50. The topology location indices, in order of the force path, are 1, 2, 3, 3, 4, 4, 5, and 5. The node with a score of 82.5 and an index of 1 is mapped to a main load-bearing role, the node with a score of 79.3 and an index of 2 is mapped to a main load-bearing role, the nodes with scores of 65.8 (index 3) and 61.4 (index 3) are mapped to redundancy check roles, the node with a score of 55.2 (index 4) is mapped to a redundancy check role, and the nodes with scores of 48.7 (index 4), 44.1 (index 5), and 39.9 (index 5) are mapped to transition anchoring roles. During implementation, the theoretical maximum tension benchmark value of the main load-bearing node under the high wind load conditions of this plateau is 45×10. 3N, after being generated by subsequent thresholds, sets a dynamic safety coefficient upper limit and an instability prevention lower limit. Redundant verification nodes trigger consistency anomaly warnings based on the duration of continuous deviations in real time, while transitionally anchored nodes capture transient impact events according to a set millisecond-level response window. This hierarchical labeling structure optimizes the priority of multi-source warning signals in field testing, significantly improving the warning processing speed of nodes with high scores and critical locations, while maintaining a lower warning frequency for edge nodes in non-critical risk events, effectively improving system resource scheduling efficiency and warning accuracy.

[0081] like Figure 3 As shown, step S4 involves extracting the theoretical maximum tension and the slope of tension change over the past ten minutes under the current working condition for the functional role label marked as the main load-bearing type, and generating a dual-modal envelope threshold range for the main load-bearing type that includes an upper limit for the dynamic safety factor and a lower limit for preventing instability. Specifically, this includes: S4.1: Based on the service status fingerprint ID of the guy wire bound to the main load-bearing functional role label, obtain the traction speed, span length and meteorological environment parameters of the current construction stage as input conditions, perform working condition mapping processing, and generate the theoretical maximum tension benchmark value that characterizes the theoretical bearing limit of the guy wire under the current working scenario.

[0082] Based on the service status fingerprint ID of the guy wire bound to the functional role label that has been marked as the main load-bearing type, the current construction stage working condition information of the corresponding monitoring point is retrieved from the remote monitoring system, and the three core input conditions of traction speed, span length and meteorological environment parameters are obtained by parsing.

[0083] By combining the traction speed value with the span length, and using the static equilibrium condition to calculate the reference axial tension component of the cable during this operation phase, the structural input of the multiphysics simulation model is used.

[0084] The wind speed and direction, temperature and humidity, and air pressure data in the meteorological environment parameters are converted into a load matrix, and the sampling time during the construction phase is matched according to the time series to generate an external disturbance input sequence.

[0085] A composite mechanical model considering the self-weight of the guy wire, dynamic traction load, temperature expansion and wind vibration is established in the simulation model, and the boundary condition constraints between the guy wire and the fixed point are realized by finite element element division.

[0086] An iterative solution strategy was adopted to calculate the maximum stress state of the pull-down cable under different combinations of external working conditions, and the peak stress value was extracted as the initial value of the theoretical bearing limit.

[0087] The initial value is matched with the installation process parameters fixed by the fingerprint ID, and the initial value is multiplied and corrected according to the material strength correction coefficient and the hardware connection efficiency coefficient to generate the theoretical maximum tension reference value.

[0088] Through the above processing method, the functional role labeling data of the previous step is transformed into a quantifiable theoretical maximum tension index under the current working condition, thereby establishing the basic conditions for subsequent dual-modal envelope threshold calculation.

[0089] S4.2: Based on the theoretical maximum tension benchmark value and the historical tension time series data window collected at the edge end, extract the tension sampling sequence within the last ten minutes as the input object, and use sliding window linear regression to perform trend fitting processing to generate the tension change slope feature quantity within the last ten minutes that reflects the transient fluctuation characteristics of the tension line under stress.

[0090] Based on the theoretical maximum tension benchmark value and historical tension time-series data windows collected at the edge, a tension sampling sequence with a time length of 600 seconds was selected as the analysis input. Timestamp resampling was performed on the sampling sequence to unify irregular sampling intervals to a fixed sampling period, forming a structured time-series matrix. Noise suppression was performed on the time-series matrix, using a preprocessing algorithm based on a combination of median filtering and detrending to eliminate high-frequency sampling noise and slow-changing trends, obtaining a stable tension signal baseline. The processed tension signal was divided into multiple continuous and partially overlapping sub-series windows using a sliding window strategy. The length and step value of each window were determined according to a preset sampling rate and analysis resolution. Linear regression fitting was performed on each sub-series window, with time as the independent variable and tension sampling value as the dependent variable, to calculate the slope of the fitted line within the window, forming a slope feature sequence. Statistical feature extraction was performed on the slope feature sequence to calculate the mean, extreme values, and directional change indices of the slope features over the past ten minutes, generating a slope feature quantity reflecting the transient fluctuation characteristics of the tension line under stress over the past ten minutes. By using sliding window linear regression, the time series data of the theoretical maximum tension benchmark value is transformed into a quantifiable slope feature, thereby achieving an accurate characterization of the rate of change of force on the tension wire.

[0091] For example, in a construction scenario with a high-voltage transmission line span of 420 meters, a traction speed of 0.35 meters per second, and a sampling frequency of 10 Hz, the historical tension time-series data window collected at the edge contains 6000 sampling points. This window is divided into 115 sub-sequence windows using a sliding window of 200 points in length and 50 points in step size. After noise suppression and detrending processing with a median filter window width of 5 points and a polynomial detrending order of 1, a stable tension signal baseline is obtained. Linear regression fitting is performed on each window, resulting in a slope characteristic value range of -0.12 to 0.15 kN / s, with a mean of 0.04 kN / s. The extreme positive peak occurs in window 87 (corresponding to time period 435 to 455 seconds), and the directional change index shows that the positive slope lasts longer than the negative slope. This slope characteristic is input into the subsequent dynamic safety factor adjustment function to generate a safety factor correction factor that adjusts with speed, significantly improving the system's early warning sensitivity and accuracy under sudden load changes.

[0092] S4.3: Based on the characteristic quantity of the slope of the tension change over the past ten minutes and the preset initial value of the safety margin, a dynamic safety factor adjustment function is constructed as the core processing mechanism, and a nonlinear gain mapping operation is performed to generate a dynamic safety factor correction factor that is adaptively adjusted with the rate of tension change.

[0093] Based on the slope characteristic of tension changes over the past ten minutes and the preset initial safety margin value, the parameter parsing interface of the dynamic threshold calculation module is called. The slope value obtained from the preceding trend fitting and the initial safety margin value are loaded as dual input vectors into the input of the nonlinear adjustment operator. The slope value is normalized and mapped to a standard interval so that it can be used in subsequent function construction under a unified dimension with the initial safety margin value. Combining the normalized slope and the initial safety margin value, a nonlinear gain function curve is constructed using polynomial expansion to achieve differentiated assignment of gain adjustment amplitude for different slope intervals. Using the function curve, the dynamic safety factor correction factor is solved. Specifically, through gain mapping operation, the normalized slope input is mapped to the corresponding safety factor adjustment ratio value, and multiplied with the initial safety margin value to generate a dynamic safety factor correction factor that adaptively adjusts with the slope change rate. The gain mapping relationship is defined using the following formula: ; Where S0 is the initial safety margin value, R is the normalized slope characteristic, k is the slope gain coefficient, α is the nonlinear adjustment exponent, and ΔS is the dynamic safety factor correction factor. Through the above formula and processing method, the slope characteristic and initial safety margin value from the previous step are transformed into a dynamic safety factor correction factor that can adapt to real-time operating condition fluctuations, achieving continuous adaptive optimization of the main load-bearing threshold range.

[0094] For example, in a high-voltage transmission line construction scenario, the theoretical maximum tension of the main load-bearing guy wire under the current working conditions is 12.5 kN. The normalized characteristic value of the tension change slope over the past ten minutes is 0.35. The initial safety margin is set to 1.8, the slope gain coefficient k is set to 0.6, and the nonlinear adjustment index α is taken as 2. Substituting the above parameters into the formula: ; First, calculate the gain term: 0.6 × 0.35 = 0.21. Adding 1 to the gain gives 1.21, which is then squared to get 1.4641. Multiplying this by the initial safety margin value of 1.8 gives ΔS ≈ 2.6354. This value is the adaptively adjusted dynamic safety factor correction factor. Applying this correction factor to calculate the upper limit of the theoretical maximum tension benchmark value, we obtain a dynamic upper limit value ≈ 32.9425 kN, a significant increase compared to the unadjusted 22.5 kN. This effectively prevents false alarms when dealing with sudden high-speed traction during construction, while ensuring a rapid response to instability risks.

[0095] S4.4: Based on the theoretical maximum tension benchmark value and the dynamic safety factor correction factor, perform multiplicative weighted fusion processing to calculate the warning upper limit. At the same time, perform reverse derivation operation based on the tension change slope characteristic of the past ten minutes to determine the instability critical point, thereby generating a main load-bearing dual-mode envelope threshold range that includes the upper limit of the dynamic safety factor and the lower limit of the instability prevention limit.

[0096] Based on the theoretical maximum tension benchmark value and the dynamic safety factor correction factor, the multiplicative weighted fusion processing module is invoked to perform numerical multiplication of the two values ​​according to the dynamic weight factor, forming a warning upper limit value representing the allowable upper limit of tension under the current working condition. The characteristic quantity of the tension change slope over the past ten minutes is input into the reverse critical deduction module to establish a trend curve of the tension reduction process. The lower limit critical value of tension is derived by analyzing the curve intersection points. A dual-modal envelope construction function is used, with the warning upper limit and the instability critical value as the upper and lower bounds of the envelope curve, respectively. The interval adaptation characteristics of the real-time tension data are fitted and corrected to make the envelope interval adaptable to changes in the construction environment. The corrected upper and lower bound data are encapsulated into a structured dual-modal envelope threshold interval parameter set. Through the above processing method, the theoretical maximum tension benchmark value and tension change slope and other characteristics from the previous step are transformed into a main load-bearing type dual-modal envelope threshold interval with a dynamic safety factor upper limit and an anti-instability lower limit, realizing a precise real-time warning boundary for the core load-bearing wire.

[0097] S4.5: Based on the generated main load-bearing dual-modal envelope threshold range, encapsulate it into a structured threshold configuration instruction as an output object to complete the hierarchical matching of the real-time tension data of the main load-bearing guy wire and the update configuration of the abnormal triggering logic.

[0098] Step S5: For functional role tags labeled as redundant verification type, select real-time tension data of main load-bearing type guy wires within the same cluster as a benchmark reference, calculate the duration of continuous deviation of the ratio of the current guy wire tension to the benchmark reference, and generate a redundancy verification type deviation consistency threshold condition for determining consistency anomalies. Specifically, this includes: S5.1: Based on the generated wire cluster topology with strong functional coupling, target nodes with main load-bearing functional role labels are selected from the set of nodes within the cluster, and the real-time tension time series data sequence of the target nodes within the current time window is extracted to construct a benchmark reference tension curve for subsequent comparative analysis.

[0099] Based on the generated wire cluster topology with strong functional coupling, the node attribute index module is called to perform functional role label filtering operation on the node set within the cluster, so as to form a target node subset that only contains nodes with the main load-bearing functional role label.

[0100] Data channel initialization is performed on the real-time tension acquisition port bound to each node in the target node subset to ensure that the timestamp accuracy of the acquired data reaches the millisecond level, so as to meet the synchronization requirements of subsequent comparison and analysis.

[0101] Using the edge tension data buffer, the tension time series data of each node in the target node subset within the current time window is read in segments. The segment length is set according to the preset time window configuration parameters, such as sampling once per second within one minute, forming an equally spaced tension sampling sequence.

[0102] The tension sampling sequence is subjected to noise reduction filtering. A bandpass filter based on fast Fourier transform is used to suppress noise components above and below a specified frequency threshold to ensure that the reference curve retains only the effective frequency band signal that reflects the actual mechanical response.

[0103] Curve fitting is performed on the filtered tension time series, and the moving average method is used to smooth short-period fluctuations, forming a benchmark tension curve with high continuity and easy comparison, so that redundant verification functional role nodes can perform consistency deviation assessment in subsequent steps.

[0104] Through the above processing method, the spectrum cluster structure of the previous step is transformed into a reference tension curve with high synchronization and low noise characteristics, so as to realize the construction of accurate input data for the determination of redundancy check ratio deviation.

[0105] For example, at a high-voltage transmission line construction site, the system identifies three guy wire nodes within a guy wire cluster, each labeled with a primary load-bearing functional role. Each node is equipped with a high-precision tension sensor with a range of 0-50kN and a sampling frequency of 1Hz, ensuring millisecond-level timestamp accuracy. After data channel initialization, the system sequentially collects 60 tension data points from each node within a 60-second monitoring window. After processing with an FFT bandpass filter (passband from 0.1Hz to 2Hz), low-frequency temperature drift and high-frequency electromagnetic interference are removed. The moving average window width is set to 5 seconds, and three continuous, smoothed reference curves are formed from the tension curve data after moving average smoothing. Assuming that the tension sequence of the first main load-bearing type tension wire after filtering remains stable in the range of 40kN±0.3kN between the 20th and 40th seconds, this curve serves as the direct reference input for the comparison of redundant verification type nodes. Through curve fitting, its smoothness is significantly improved, noise components are completely eliminated, and the final output benchmark reference curve can achieve millisecond-level synchronization and significantly improve the matching degree in subsequent consistency deviation calculations.

[0106] S5.2: Obtain the real-time collected tension value of the monitored tension node with redundant verification function role label, and use the synchronous clock signal to align the real-time collected tension value with the aforementioned benchmark tension curve on the time axis to generate a dual-channel tension comparison dataset containing synchronous timestamps.

[0107] For the tension cable nodes to be monitored with redundant verification function role tags, the tension acquisition channel of the edge MCU is invoked to obtain the real-time acquired tension value sequence within the current time window, and data integrity checks are performed to eliminate missing or outliers. Based on the reference tension curve of the same cluster of main load-bearing tension cables constructed in the previous step S5.1, the internal clock module is invoked to generate a unified synchronization clock signal to ensure millisecond-level consistency of timestamps from different acquisition sources. Time axis mapping operations are performed to cross-match the timestamp sequence of the tension cable to be monitored with the timestamp sequence of the reference tension curve to locate the data point index at the corresponding time position in each sequence. A dual-channel tension comparison buffer is constructed, storing the matched tension data to be monitored and the reference tension data in independent channels, and binding a unified synchronization timestamp tag in the buffer. Data structure rearrangement is performed to form a comparison set in the form of triplets for the dual-channel tension data and the unified timestamp, realizing the generation of a complete dual-channel tension comparison dataset. Through the above synchronization alignment processing, the reference curve and the tension sequence to be monitored in the previous step are transformed into a comparison set containing time consistency identifiers to ensure strict time sequence matching in subsequent ratio calculations.

[0108] S5.3: Perform point-by-point division on the tension values ​​of the tension to be monitored and the reference tension values ​​at the same moment in the dual-path tension comparison dataset to calculate the instantaneous tension ratio sequence that characterizes the degree of matching of the mechanical responses of the two.

[0109] The system acquires a dual-channel tension comparison dataset containing synchronization timestamps as input. The data sequence parsing module extracts the corresponding data points between the real-time tension value of the guy wire to be monitored and the reference tension value. A numerical matching verification operation is performed to ensure a one-to-one correspondence between the two data streams at the same time point to meet the requirements of point-by-point calculation. The floating-point division unit is then invoked to divide the tension value of the guy wire to be monitored by the reference tension value at each time point, forming an instantaneous ratio calculation formula to characterize the degree of mechanical response matching.

[0110] The calculation results are written to the ratio sequence buffer, and a corresponding timestamp is appended to each ratio data point to maintain temporal consistency. Precision control processing is performed on the generated instantaneous tension ratio sequence, converting floating-point results to a preset number of decimal places to ensure numerical stability in subsequent threshold comparisons. Through this processing, the dual-path tension comparison data from the previous step is transformed into an instantaneous ratio sequence capable of quantifying the degree of matching in their mechanical responses, enabling fine-grained evaluation of the force synchronization between redundant verification guy wires and main load-bearing guy wires.

[0111] S5.4: Set a preset upper and lower limit for the normal fluctuation range of the tension ratio, logically compare each data point in the instantaneous tension ratio sequence with the normal fluctuation range of the tension ratio, identify abnormal ratio data points that exceed the normal fluctuation range of the tension ratio, and count the number of consecutive occurrences of the abnormal ratio data points on the time axis to generate a parameter for the continuous deviation duration of the tension ratio.

[0112] The input objects are the instantaneous tension ratio sequence output from step S5.3, and the pre-set upper and lower limits of the normal fluctuation range of the tension ratio. This sub-step, in the redundant verification threshold condition judgment chain, undertakes the function of converting the instantaneous ratio into a continuous deviation duration parameter, so as to realize the quantitative judgment basis for the abnormal consistency of the tension wire mechanical response.

[0113] For each data point in the instantaneous tension ratio sequence, perform an upper and lower limit interval logical comparison, and mark the result of the ratio being greater than the upper limit or less than the lower limit as an abnormal state bit.

[0114] The abnormal status bits are arranged in timestamp order. By using continuous region identification, segments with true abnormal status bits at adjacent time points are merged into a single continuous abnormal segment.

[0115] For each continuous abnormal section, perform length statistics calculation, convert the length value into a time unit, and obtain the corresponding tension ratio continuous deviation duration.

[0116] The duration of all consecutive abnormal segments is stored as a sequence, and the maximum value is selected as the continuous deviation time parameter of the tension ratio within that time window.

[0117] In the formula calculation, let the instantaneous tension ratio be R. i If the lower limit of normal fluctuation is L and the upper limit of normal fluctuation is U, then the logic for anomaly detection is as follows: ; The symbol ∨ represents the logical OR operation. The formula for calculating the duration of abnormal events is: ; Where Δt is a single sampling time interval, and n is the number of consecutive abnormal state positions. Through the above chain derivation process, the ratio sequence of the previous step is transformed into a tension ratio continuous deviation duration parameter, realizing the generation of the core input for the consistency anomaly triggering condition.

[0118] S5.5: Compare the continuous deviation duration parameter of the tension ratio with the preset minimum continuous deviation time threshold. When the continuous deviation duration parameter of the tension ratio is greater than or equal to the minimum continuous deviation time threshold, trigger the consistency anomaly judgment logic and generate a redundant verification type deviation consistency threshold condition for activating the graded early warning signal.

[0119] Step S6: For functional role labels marked as transitionally anchored, a millisecond-level response window is set to capture the rate of tension increase, and combined with the steady-state tension tolerance parameter, a gradient-sensitive threshold triggering rule sensitive to transient impacts is generated. Specifically, this includes: S6.1: Obtain the real-time tension sampling sequence and historical steady-state tension distribution characteristics of the transition anchoring type guy wire, and perform dispersion analysis on the historical steady-state tension distribution characteristics based on sliding window statistics to generate a steady-state tension tolerance baseline parameter that characterizes the normal fluctuation range of the guy wire in this role.

[0120] S6.2: Using the steady-state tension tolerance baseline parameter as a dynamic filtering reference, perform first-order difference operation processing on the real-time tension sampling sequence to extract the instantaneous tension change rate vector that reflects the degree of force change per unit time.

[0121] Using the steady-state tension tolerance baseline parameter as a reference standard for real-time filtering operations, dynamic feature constraint processing is applied to the input real-time tension sampling sequence of the transition anchoring type guy wire. The steady-state tension tolerance baseline parameter is loaded into the amplitude threshold register of the differential operation control module, setting the maximum allowable fluctuation range of the differential result. Continuous sampling data blocks of the real-time tension sampling sequence are retrieved, and the amplitude difference between adjacent sampling points is calculated. A preliminary rate of change sequence reflecting the rate of force change per unit time is generated through a differential operator. The preliminary rate of change sequence is compared with the steady-state tension tolerance baseline parameter in amplitude, and the sensitivity of the differential result is adjusted using an adaptive scaling factor to eliminate low-amplitude and noisy rate of change signals. High-frequency noise suppression filtering is applied to the adjusted rate of change sequence, and a window function weighted averaging method is used to smooth abrupt changes, ensuring the stability of the output rate of change vector and the significance of the impact response characteristics. Through this dynamic filtering and first-order differential chain operation, the real-time tension sampling sequence is transformed into an instantaneous tension rate of change vector, which is then used in the subsequent millisecond-level response window peak search stage to achieve precise and sensitive capture of transient mechanical impacts.

[0122] S6.3: Based on a preset millisecond-level high-frequency response time window, perform peak search and duration determination processing on the instantaneous tension change rate vector to filter out candidate impact event segments that exceed the normal fluctuation frequency and have transient impact characteristics.

[0123] Based on a preset millisecond-level high-frequency response time window, the instantaneous tension change rate vector output in the preceding step S6.2 is used as the high-frequency detection signal input. The instantaneous tension change rate vector is processed by window segmentation, with each window length equal to the preset millisecond-level response time parameter, and the overlap ratio of sampling points between adjacent windows is strictly maintained to enhance peak capture accuracy. Extreme value extraction is performed on the change rate vector within each window, calculating the maximum rising change rate value and the corresponding occurrence time point for each window segment, and outputting it as a peak candidate list. Duration analysis is performed on each element in the peak candidate list, comparing the length of the continuous high change rate segment before and after the peak with a preset minimum impact duration threshold to filter out short noise events that do not meet the duration condition, while recording the start and end times of segments that meet the condition.

[0124] The frequency characteristics of candidate events are calculated using a high-frequency fluctuation determination formula: ; Where f is the frequency characteristic value of the impact event, and Δt is the time interval between adjacent peaks. The calculated frequency characteristic value is compared with a preset upper limit of normal fluctuation frequency to identify events that exceed the normal fluctuation frequency range. For events that meet both the frequency exceeding the limit and the duration requirement, feature annotation is performed and the results are output as a candidate impact event fragment dataset.

[0125] Through the above processing method, the instantaneous rate of change vector of the previous step is transformed into candidate impact event fragments with peak amplitude, duration and frequency characteristics, so as to achieve the screening of real transient impact characteristics of transition anchoring type guy wires on the millisecond time scale.

[0126] For example, in a construction scenario of erecting a high-voltage transmission tower, a millisecond-level response window width of 5ms is set on the transition anchoring guy wire. The instantaneous tension change rate vector is output by a tension sensor with a real-time sampling frequency of 1kHz, and the window overlap ratio is set to 50%. The maximum rate of change value of each window segment is calculated to yield 5 peak candidates, of which 3 events have durations of 12ms, 15ms, and 18ms, respectively, exceeding the preset minimum impact duration threshold of 10ms. In the frequency characteristic calculation formula, the frequency characteristic values ​​are 25Hz, 33.33Hz, and 50Hz, respectively, obtained by taking the time interval Δt between adjacent peaks as 0.04s, 0.03s, and 0.02s. All of these are greater than the upper limit of the normal fluctuation frequency of 20Hz, and are judged as abnormal transient impact events. The output candidate impact event fragment dataset is used for multi-level threshold discrimination in the next sub-step S6.4. The verification results show that the system can identify high-amplitude impact events within 10ms after the impact occurs, significantly improving the ability to capture transient mechanical impacts.

[0127] S6.4: Based on the rising edge slope value and duration index of the candidate impact event segment, and combined with the reverse constraint logic of the steady-state tension tolerance baseline parameter, perform multi-level threshold comparison operation to generate a gradient-sensitive discrimination flag that distinguishes between real mechanical impact and noise interference.

[0128] The rising edge slope and duration of candidate impact event segments are obtained as inputs for the decision.

[0129] Threshold segmentation mapping is performed on the rising edge slope value to divide different amplitude change rates into three levels: high, medium, and low, and a corresponding comparison threshold is set for each level.

[0130] The duration index is processed into intervals, mapping the time span to three segments: short time domain, medium time domain, and long time domain, and associating threshold conditions for each time domain.

[0131] Input the steady-state tension tolerance baseline parameter into the reverse constraint logic module to generate a reverse tolerance range for anomaly judgment, so as to avoid misjudging steady-state fluctuations as shocks.

[0132] Perform multi-level threshold comparison operations, and compare the rising edge slope level, duration level and reverse tolerance interval in combination to determine the impact effectiveness corresponding to each combination.

[0133] Gradient-sensitive discrimination flags are generated based on the combined comparison results, and the flags are matched and verified with the noise interference pattern library to remove flags that are consistent with known noise patterns.

[0134] By comparing multiple thresholds and matching with a noise pattern library, the candidate impact event fragments from the previous step are transformed into validated gradient-sensitive discrimination flags, thus enabling effective differentiation between real mechanical impacts and noise interference.

[0135] S6.5: Based on the triggering state of the gradient sensitivity discrimination flag, dynamically bind the corresponding millisecond-level response window width and steady-state tension tolerance relaxation coefficient to finally generate a gradient sensitivity threshold triggering rule configuration set that adapts to the transition anchoring role.

[0136] Step S7: The real-time collected tension data is input into the corresponding functional role tag-bound main load-bearing dual-modal envelope threshold range, redundancy check type deviation consistency threshold condition, or gradient sensitive threshold triggering rule for matching and judgment, so as to generate a graded early warning trigger signal. Specifically, this includes: S7.1: Obtain the real-time tension sampling sequence and the threshold range of the main load-bearing dual-mode envelope, and perform upper and lower limit envelope squeezing judgment processing on the real-time tension sampling sequence to generate the main load-bearing over-limit state flag.

[0137] The real-time tension sampling sequence bound to the main load-bearing functional role tag is obtained as the input object. The structured tension value set output by the edge data acquisition module is called, and the bimodal envelope threshold interval parameters generated in the current period are loaded. The upper and lower limits of the bimodal envelope threshold interval are mapped to the corresponding physical constraints, and associated with the current working condition dynamic safety factor and the inversion parameter of the anti-instability lower limit, respectively, to form a numerical interval matrix that can be directly used for judgment. For each sampling point in the real-time tension sampling sequence, the interval squeeze judgment method is used, and the squeeze state is calculated using the following formula: ; Where T is the current sampled tension value, T upper T is the upper limit of the dynamic safety factor. lower To prevent instability at the lower limit, Boolean aggregation is performed on the judgment results of all sampling points. The proportion of sampling points that meet the over-limit condition is compared with a preset trigger proportion threshold to calculate the proportion judgment value. Based on the relationship between the proportion judgment value and the trigger proportion threshold, a flag bit for the main load-bearing type over-limit state is generated. This flag bit serves as the trigger condition for subsequent redundant verification logic. Through the above squeeze judgment and proportion trigger processing method, the real-time tension sampling data from the previous step is transformed into a clear over-limit state indication, achieving accurate initial screening of abnormal forces on the main load-bearing type.

[0138] S7.2: Obtain the main load-bearing type over-limit status flag and the same cluster reference tension data. Based on the main load-bearing type over-limit status flag, trigger the redundancy verification logic, calculate the duration of continuous deviation of the ratio of the current tension to the reference tension data, and generate a redundancy verification type consistency anomaly judgment result.

[0139] The main load-bearing type over-limit status flag and the same cluster reference tension data are obtained as the trigger condition input for the redundancy verification logic.

[0140] The real-time tension sampling value of the current monitoring cable is synchronized with the benchmark tension curve to ensure that the two data streams can be compared and analyzed under a unified time axis.

[0141] Perform point-by-point ratio calculations on the synchronized data pairs to form an instantaneous tension ratio sequence.

[0142] The instantaneous tension ratio sequence is compared point by point with the upper and lower boundaries of the preset normal fluctuation range to identify all abnormal ratio points that exceed the range, and the duration of continuous occurrence of these abnormal points is recorded.

[0143] The recorded duration of continuous deviation is compared with the set minimum continuous deviation time threshold. When the deviation duration meets or exceeds the threshold, a redundancy check type consistency anomaly judgment result is output.

[0144] By quantitatively calculating the duration of continuous deviation, the redundant verification results associated with the main load-bearing type over-limit events are transformed into consistency risk judgment data, enabling timely identification and graded early warning of abnormal tension distribution within the cluster.

[0145] S7.3: Obtain the time differential features of the real-time tension sampling sequence and the millisecond-level response window parameters in the gradient sensitive threshold triggering rule, and perform sliding window rate extreme value extraction processing on the time differential features to generate a transition anchoring transient impact capture signal.

[0146] S7.4: Obtain the main load-bearing type over-limit status flag, the redundancy check type consistency anomaly judgment result, and the transition anchoring type transient impact capture signal. Perform multi-source signal weighted fusion processing based on the preset role priority mapping table to generate the initial hierarchical early warning coding vector.

[0147] The main load-bearing type over-limit status flag, the redundancy check type consistency anomaly judgment result, and the transition anchoring type transient impact capture signal are obtained as multi-source input vectors. The weight coefficient matrix set for different functional roles in the role priority mapping table is called, and each input vector is multiplied element by element with the corresponding weight coefficient to generate a weighted component matrix.

[0148] Based on the weighted component matrix, column vector summation is performed to generate a sequence of aggregated values ​​for the role components. A normalization transformation is then applied to this sequence to eliminate differences in numerical dimensions, ensuring that the contributions of different input sources in the fusion process are comparable.

[0149] A Gaussian-shaped smoothing kernel function is used to perform convolution operations on the normalized sequence to eliminate instantaneous noise interference and preserve trend changes, outputting a smoothed fused trend sequence.

[0150] A threshold segmentation mapping function is applied to the fusion trend sequence to map fusion signals below the first segment threshold to low-level early warning codes, fusion signals between the first and second segment thresholds to medium-level early warning codes, and fusion signals above the second segment threshold to high-level early warning codes, in order to generate an initial hierarchical early warning code vector.

[0151] Through the above-mentioned multi-source signal weighted fusion and segmented mapping processing, the result of the previous step is transformed into an early warning code with grade labels, realizing unified evaluation and response ranking of abnormal signals from multiple roles.

[0152] For example, in the guy wire monitoring of a 220kV transmission line erection project, the value of the main load-bearing type over-limit status flag is set to 1, the value of the redundancy check type consistency anomaly judgment result is set to 0.6, and the value of the transition anchoring type transient impact capture signal is set to 0.3. The weight coefficient matrix in the role priority mapping table is configured as 0.5 for main load-bearing type, 0.3 for redundancy check type, and 0.2 for transition anchoring type. After performing element-wise multiplication on the three types of inputs, the component matrix {0.5, 0.18, 0.06} is obtained. The summation yields an aggregate value of 0.74, which is normalized to 0.74. After processing with a smoothing kernel function, the result remains 0.74. The first segment threshold is set to 0.5, and the second segment threshold is set to 0.8. Since the fused value is between the two, it corresponds to the intermediate warning coding vector [0, 1, 0]. In actual operation, this coding vector drives the intermediate alarm push, allowing construction personnel to check the site nearby and adjust the guy wire tension within 90 seconds. This significantly improves response efficiency and risk handling priority compared to the traditional unified threshold strategy.

[0153] S7.5: Obtain the confidence correction coefficients corresponding to the initial graded early warning coding vector and the equipment health decay curve, and perform dynamic weight normalization calibration on the initial graded early warning coding vector to generate the final graded early warning trigger signal.

[0154] Step S8: Based on the false alarm or missed alarm feedback records of the graded early warning trigger signal, perform closed-loop fine-tuning of the dynamic safety coefficient, ratio deviation duration, or response window parameters on which the signal is generated, to update the role dynamic range parameter configuration for the next cycle. Specifically, this includes: S8.1: Obtain the graded early warning trigger signals generated within the historical period and their corresponding actual working condition verification results, and perform consistency comparison processing on the graded early warning trigger signals and actual working condition verification results to generate an early warning effectiveness evaluation dataset containing false alarm event identifiers and missed alarm event identifiers.

[0155] S8.2: Based on the false alarm event identifiers in the aforementioned early warning effectiveness evaluation dataset, extract the associated functional role labels and the dynamic safety coefficient, ratio deviation duration, or response window parameters that were in effect at the time. Perform reverse causal deduction processing on the parameters that were in effect at the time and the tension time sequence characteristics when the false alarm occurred to generate the parameter sensitivity deviation vector that caused the false alarm.

[0156] S8.3: Based on the missed event identifiers in the early warning effectiveness evaluation dataset, extract the untriggered potential risk segments and the current role adaptive interval boundary, and perform coverage loss analysis on the tension change gradient between the role adaptive interval boundary and the potential risk segments to generate a threshold hysteresis compensation amount characterizing the range of the early warning blind zone.

[0157] Before performing the missed detection compensation analysis, the input objects are the event records marked as missed detections in the early warning effectiveness evaluation dataset, including the associated potential risk segment tension time series curves and the currently effective role adaptive interval boundary parameters. Time differentiation is performed on the potential risk segment tension time series curves to extract gradient vectors reflecting unit-time tension changes, serving as the basic data for coverage analysis. The gradient vectors are then used to perform point-by-point interval inclusion determination with the adaptive interval boundary values ​​bound to the corresponding functional role labels, generating a binary label sequence distinguishing covered and uncovered risk points. Statistical analysis is performed on the gradient vector set of uncovered risk points, calculating their mean and peak gradients as characteristic indicators of the risk segments. A difference function is used to calculate the deviation between the risk segment gradient characteristics and the current interval boundary threshold. Positive value filtering is applied to the difference to remove covered risk points, retaining only the missed detection portion caused by insufficient threshold. The maximum deviation and mean deviation are selected from the positive value filtering results, and combined with the risk segment duration indicator, a threshold lag compensation amount representing the range of the early warning blind zone is generated through a weighted summation. Through the above gradient difference statistics and weight fusion processing, the risk segment characteristics in the missed events are transformed into threshold lag compensation quantities that can be directly used for parameter correction, thereby achieving targeted optimization of the warning interval boundary for the next cycle.

[0158] S8.4: Using the parameter sensitivity deviation vector and threshold hysteresis compensation, construct a parameter optimization function under multi-objective constraints, and perform iterative correction processing on the dynamic safety factor of the current main load-bearing dual-modal envelope threshold interval, the ratio deviation time of the redundant verification type deviation consistency threshold condition, and the response window parameters of the transition anchoring type gradient sensitive threshold triggering rule, so as to generate an updated role adaptive optimization parameter set.

[0159] Based on the generated parameter sensitivity deviation vector and threshold hysteresis compensation, a parameter optimization function under multi-objective constraints is constructed. The input conditions are the dynamic safety coefficient values ​​bound to the functional role labels of each period, the redundancy check ratio deviation duration values, and the transition anchoring response window width values. The parameter sensitivity deviation vector is decomposed into single-role deviation components for different functional roles, and the threshold hysteresis compensation is mapped to the time scale components of each role in the analysis of the lack of coverage in the early warning blind zone, serving as the constraint set for the optimization function. This constraint set is used to initialize the parameter optimization function, enabling it to perform joint optimization in a multi-role, multi-index space. The objective terms of the optimization function are defined in a structured manner, where the main load-bearing objective term is to minimize the mean square error of the dynamic safety coefficient relative to the theoretical safety coefficient, the redundancy check objective term is to minimize the absolute value of the difference between the ratio deviation duration and the coverage rate of the normal fluctuation range, and the transition anchoring objective term is to minimize the difference between the impact event capture rate and the noise misjudgment rate in the response window width. A multi-objective weighted summation strategy is adopted, aggregating each objective item according to the weight coefficients of the role risk level mapping table to form a comprehensive evaluation index. Gradient descent iterative solution is used to progressively update each parameter within the optimization function value space. Each iteration depends on the directional adjustment of the parameter values ​​updated in the previous round and the current deviation component, ensuring that the parameters converge to an optimization interval that simultaneously satisfies the multi-objective constraints. During the iteration process, when the decrease in the comprehensive evaluation index falls below the preset convergence threshold or reaches the maximum number of iterations, the iteration terminates and the updated dynamic safety coefficient, ratio deviation duration, and response window width are output, constituting the updated role adaptive optimization parameter set. Through the above multi-objective constraint iterative optimization processing method, the deviation vector and compensation results of the previous step are transformed into precise parameters that can be directly applied to the threshold generation logic of the next cycle, realizing scientific closed-loop optimization of the role adaptive interval in a scenario of dynamic adjustment of multi-device warning priorities.

[0160] S8.5: Write the updated role adaptive optimization parameter set into the local non-volatile storage unit of the monitoring device and synchronize it to the attribute field of the corresponding node in the cloud fingerprint evolution map. Perform hot replacement processing on the threshold generation logic of the next monitoring cycle to complete the closed-loop update of the role dynamic range parameter configuration.

[0161] For those skilled in the art, various other corresponding changes and modifications can be made based on the technical solutions and concepts described above, and all such changes and modifications should fall within the protection scope of the claims of this invention.

[0162] Unless otherwise defined, the technical or scientific terms used herein shall have the ordinary meaning as understood by one of ordinary skill in the art to which this application pertains. The terms “first,” “second,” “third,” and similar terms used in this patent application specification and claims do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Similarly, the terms “an” or “a” and similar terms do not indicate a quantity limitation, but rather indicate the presence of at least one. The terms “comprising” or “including” and similar terms mean that the elements or objects preceding “comprising” or “including” encompass the elements or objects listed following “comprising” or “including” and their equivalents, and do not exclude other elements or objects. The “multiple” mentioned in the embodiments of this application refers to two or more. A and / or B indicate three possibilities: A; B; and A and B.

[0163] The above description is merely an exemplary embodiment of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and such modifications or substitutions should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A remote intelligent monitoring method for guy wire stress in power grid construction, specifically including: S1: Obtain the four types of initial fingerprint elements solidified during the installation phase of the wire monitoring device to generate the wire service status fingerprint ID; S2: Construct a fingerprint evolution graph, map the service status fingerprint ID of the pull wire to the fingerprint evolution graph node, calculate the cross-dimensional similarity between different nodes, and generate graph edge weights; S3: Utilize graph edge weights to perform subgraph aggregation on the fingerprint evolution graph, identify wire clusters, and combine topological location, historical tension peak distribution dispersion and redundancy check success rate indicators to dynamically label each node in the cluster with functional role tags of main load type, redundancy check type or transition anchoring type. S4: For functional role labels marked as main load-bearing type, extract the theoretical maximum tension and the slope of tension change in the past ten minutes under the current working condition, and generate the dual-modal envelope threshold range of main load-bearing type. S5: For functional role tags labeled as redundant verification type, select the real-time tension data of the main load-bearing type guy wire in the same cluster as the benchmark reference, calculate the duration of continuous deviation of the ratio of the current guy wire tension to the benchmark reference, and generate the redundancy verification type deviation consistency threshold condition. S6: For functional role labels marked as transition anchoring type, set a millisecond-level response window to capture the tension rise rate, and combine it with the steady-state tension tolerance parameter to generate gradient sensitive threshold triggering rules. S7: Input the tension data into the main load-bearing dual-modal envelope threshold range, the redundancy check type deviation consistency threshold condition, or the gradient sensitive threshold triggering rule bound to the corresponding functional role label for matching and judgment, and generate a graded early warning trigger signal.

2. The remote intelligent monitoring method for guy wire stress in power grid construction according to claim 1, characterized in that, The four types of initial fingerprint elements include installation process parameters, environmental exposure trajectory, tension timing characteristics, and equipment health decay curve.

3. The remote intelligent monitoring method for guy wire stress in power grid construction according to claim 1, characterized in that, The edge weights of the graph represent the functional coupling relationship between the lines.

4. The remote intelligent monitoring method for guy wire stress in power grid construction according to claim 1, characterized in that, Following S7, the following also includes: S8: Based on the false alarm or missed alarm feedback record of the graded early warning trigger signal, perform closed-loop fine-tuning operation on the dynamic safety coefficient, ratio deviation duration or response window parameter on which the signal is generated, so as to update the role dynamic range parameter configuration for the next cycle.

5. The remote intelligent monitoring method for guy wire stress in power grid construction according to claim 1, characterized in that, S3 specifically includes: Obtain the set of graph edge weights that characterize the functional coupling relationship between the wires, and perform iterative cutting processing on the fingerprint evolution graph to generate an initial set of wire clusters with strong internal connectivity and weak external correlation. Extract the construction traction process topology location data of each node in the initial guy wire cluster set, and perform connectivity component analysis processing in combination with the physical connection adjacency matrix between members within the cluster to generate a topology hierarchy index that reflects the relative importance of the guy wire in the force transmission path; The historical tension peak distribution dispersion statistics and redundancy check success rate records of each initial tension wire cluster are retrieved, and multi-dimensional feature fusion calculation is performed on the topology hierarchical index to generate a comprehensive performance score. Based on the ranking of comprehensive performance scores, a hierarchical mapping matching process is performed on the nodes in the initial wire cluster set. Nodes with high scores and located in key topological positions are mapped to main load-bearing functional roles, nodes with medium scores and comparable value are mapped to redundancy verification functional roles, and nodes with low scores and located at the edge are mapped to transitional anchoring functional roles, so as to generate a dynamic wire cluster structure with functional role labels.

6. The remote intelligent monitoring method for guy wire stress in power grid construction according to claim 5, characterized in that, The comprehensive performance score characterizes the stability and reliability within the cluster.

7. The remote intelligent monitoring method for guy wire stress in power grid construction according to claim 1, characterized in that, S4 specifically includes: Based on the service status fingerprint ID of the guy wire bound to the main load-bearing functional role label, the traction speed, span length and meteorological environment parameters of the current construction stage are obtained as input conditions, and the working condition mapping process is performed to generate the theoretical maximum tension benchmark value that characterizes the theoretical bearing limit of the guy wire under the current working scenario. Based on the theoretical maximum tension benchmark value and the historical tension time series data window collected at the edge end, the tension sampling sequence within the last ten minutes is extracted as the input object, and trend fitting processing is performed to generate the tension change slope feature quantity within the last ten minutes that reflects the transient fluctuation characteristics of the tension line under stress. Based on the characteristic quantity of the slope of tension change over the past ten minutes and the initial value of the safety margin, a dynamic safety factor adjustment function is constructed as the core processing mechanism to perform nonlinear gain mapping operation in order to generate a dynamic safety factor correction factor that is adaptively adjusted with the rate of tension change. Based on the theoretical maximum tension benchmark value and the dynamic safety factor correction factor, a multiplicative weighted fusion process is performed to calculate the warning upper limit. At the same time, a reverse derivation operation is performed based on the tension change slope characteristic of the past ten minutes to determine the instability critical point, thereby generating the main load-bearing dual-mode envelope threshold range. Based on the generated primary load-bearing dual-modal envelope threshold range, it is encapsulated as a structured threshold configuration instruction as an output object to complete the hierarchical matching of the real-time tension data of the primary load-bearing guy wire and the update configuration of the abnormal triggering logic.

8. The remote intelligent monitoring method for guy wire stress in power grid construction according to claim 7, characterized in that, The main load-bearing dual-modal envelope threshold range includes an upper limit for the dynamic safety factor and a lower limit for preventing instability.

9. The remote intelligent monitoring method for guy wire stress in power grid construction according to claim 1, characterized in that, S5 specifically includes: Based on the generated wire cluster topology with strong functional coupling, target nodes with main load-bearing functional role labels are selected from the set of nodes within the cluster, and the real-time tension time series data sequence of the target nodes within the current time window is extracted to construct a benchmark reference tension curve for subsequent comparative analysis. The real-time tension values ​​of the monitored tension nodes with redundant verification function role labels are obtained. The real-time tension values ​​are aligned with the aforementioned reference tension curve using a synchronous clock signal to generate a dual-channel tension comparison dataset containing a synchronous timestamp. A point-by-point division operation is performed on the tension value of the tension line to be monitored and the reference tension value at the same moment in the dual-path tension comparison dataset to calculate the instantaneous tension ratio sequence. Set the upper and lower limits of the normal fluctuation range of the tension ratio, logically compare each data point in the instantaneous tension ratio sequence with the normal fluctuation range of the tension ratio, identify abnormal ratio data points that exceed the normal fluctuation range of the tension ratio, and count the number of consecutive occurrences of abnormal ratio data points on the time axis to generate a parameter for the continuous deviation duration of the tension ratio. The continuous deviation duration parameter of the tension ratio is compared with the preset minimum continuous deviation time threshold. When the continuous deviation duration parameter of the tension ratio is greater than or equal to the minimum continuous deviation time threshold, the consistency anomaly judgment logic is triggered, and a redundant verification type deviation consistency threshold condition is generated to activate the graded early warning signal.

10. The remote intelligent monitoring method for guy wire stress in power grid construction according to claim 9, characterized in that, The instantaneous tension ratio sequence characterizes the degree of matching between the two mechanical responses.

Images